You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Some use cases dictate the parsing of raw protobuf from a topic - ie, a topic that does not have the magic byte, or the 4 byte schema id. It should be possible to add efficient DH support for raw parsing based on a Descriptor:
protobuf_spec(
...
schema_message_name="com.example.MyMessage",
type="local"# or "serdes" (the default), or "registry"
)
serdes would be the default, and would mean the schema exists on the registry and the topic is the "serdes" protocol local would mean the schema exists on the classpath and the topic is a "raw" protocol registry would mean the schema exists on the registry and the topic is a "raw" protocol
The text was updated successfully, but these errors were encountered:
It might also be reasonable to add "local" support, or point-in-time "registry" support, even if the protocol is really serdes. This might allow DH to be more efficient and skip the confluent io.confluent.kafka.serializers.protobuf.KafkaProtobufDeserializer impl.
The existing kafka protobuf feature relies on the schema registry and the kafka protobuf serdes protocol.
Some use cases dictate the parsing of raw protobuf from a topic - ie, a topic that does not have the magic byte, or the 4 byte schema id. It should be possible to add efficient DH support for raw parsing based on a Descriptor:
serdes
would be the default, and would mean the schema exists on the registry and the topic is the "serdes" protocollocal
would mean the schema exists on the classpath and the topic is a "raw" protocolregistry
would mean the schema exists on the registry and the topic is a "raw" protocolThe text was updated successfully, but these errors were encountered: